Exporting Data from an MBSE Model to a RAN Simulator : Designing and Implementing an Interface between MagicDraw and Matlab

Abstract

This thesis addresses a common challenge in adopting a model-based systems engineering (MBSE) approach: integrating the modelling tool with external tools. In this case the problem is synchronizing system-on-chip (SoC) use case scenario configurations between a MagicDraw-based MBSE model and Matlab-based 5G RAN simulator. Use cases are derived from system requirements and they are validated by simulating the scenarios of each of the use cases. Manually maintaining the scenario configurations in the simulator’s environment is time consuming and prone to human errors – an issue this work aims to mitigate. A secondary research question examined the available methods for exporting and importing data from/to MagicDraw. A review of built-in methods, third-party solutions and MagicDraw’s OpenAPI supported the selection of a custom plugin as the most flexible and maintainable approach for the implementation.

A proof-of-concept interface was developed using a custom MagicDraw plugin together with the simulator’s built-in functionality for generating simulation scripts from a spreadsheet. The plugin exports the scenario configurations into a .xls file, which is read by the simulator together with a base script to generate executable simulation scripts. This reduces the need to manually edit the scripts after changes are made to the use cases, leaving only the base script to be maintained manually.

The solution is implemented for MagicDraw, meaning it will not work for other modelling tools. It also currently supports only a few of the downlink channels. The implementation creates a foundation for future work, such as further development to support additional downlink channels or integrating other tools into the MagicDraw MBSE workflow. Although the solution has several aspects that can be further improved, it is successful as a proof-of-concept interface.

The interface was not tested with real use cases, as creating simulation-ready configurations requires product-specific parameter expertise. Instead, the solution and its features were manually tested by exporting a variety of configurations and focusing on corner cases such as missing, negative, or zero values in unexpected places. The simulator’s built-in function was able to generate script variants from the exported .xls spreadsheet. However, the base script was not fully developed during the thesis due to the detailed parameter expertise required. As a result, the generated script variants may not run end-to-end in the simulator, even though the interface itself was technically verified to function as intended